[OLPC-devel] USB device support

[John R.]

On 8/25/06, David Zeuthen <davidz at redhat.com> wrote:
...
> So, I don't really buy that user space drivers are easier than kernel
> space drivers. It's more of a myth.
>
> Of course, user space drivers are nice if you care proprietary software
> (since there is no kernel ABI that breaks all the time) and stability
> (since it only kills the user space program using it) but I guess this
> is not what you were getting at. And both niceties are totally void IMO
> for OLPC anyway: we should not care about proprietary closed drivers in
> OLPC; and we shouldn't include drivers that crash be they either user or
> kernel land. That's what I personally think, anyway.
>

Clearly you're thinking of a OLPC in the centralized/managed model
that is going to be the norm for most units.

I don't think the issue is "proprietary or not," ABI breakage, etc...
it's an issue of your hardware not currently being on the OLPC managed
service officially supported list. When that happens (which it will)
what are the user's options? Does she have to rebuild her kernel with
a specific config or compiler parameters? If there are user mode
drivers available it doesn't matter what the managed services company
decides to support. It doesn't really matter if you're willing to go
your own way by using your own kernel, but that's probably not a route
the majority of users stuck in an "unsupported hardware" bind will
take.

All that said I doubt there's anything actionable for the project
either way. The plain reality is most usb devices are supported by
kernel mode drivers not user land and there's nothing any quantity of
hot air can do about that.

-- John.